Harriry AnD RamaceE—Banded Flame-Spectra of Metals. 341 
principal lines in their spark-spectra, and merge into bands in the flame-spectra, 
or broaden when the quantity of substance is large. This is a feature which has 
already been observed in these and other spectra.* 
Copper lines, wave-lengths, : : . 3274 and 3248. 
Silver lines, wave-lengths, . c 6 3383 and 3281. 
There are no lines of gold corresponding to these in the same region, but one 
line of still higher refrangibility has been photographed, at wave-length 2675. 
The bands in the spectrum of gold are distinguished by being more refrangible 
than those which apparently correspond to them in the spectrum of copper; and we 
should, from this fact, and from the apparent homology of the spectra, expect the 
doublet of gold to lie in the same region of more refrangible rays. The line photo- 
graphed is much weaker than the corresponding lines of silver and copper. We 
attribute this fact to the insufficiency of the energy of the flame to produce these 
more rapid vibrations with the same amplitude as in the less rapid vibrations of the 
copper and silver lines. The same want of energy is apparent in the spark- 
spectrum of gold.+ 
All of these five lines have an intensity of ten in the are-spectra of the metals, 
and the second line of the gold doublet, wave-length 2428, not yet observed in the 
flame-spectrum, has also the maximum intensity. 
The flame-spectrum of gold chloride, investigated by Mitscherlich, Lecocq de 
Boisbaudran, and Damargay, does not appear to be related to the spectrum of the 
metal, 
Spectra of the Metals of the Alkaline Earths. 
Many bands are present in the oxyhydrogen flame-spectra of these metals. 
These bands, which differ entirely in character from those just described, have 
always been attributed to the oxides or to other compounds of the metals. They 
are diffuse and not degraded; neither are they composed of lines. We have no 
direct evidence of the compounds from which they are produced being dissociated 
in the flame, as is the case with the alkali metals, and we know that different 
compounds, when care is exercised that they shall not be converted into oxides or 
undergo dissociation, yield different band-spectra, though the metal be the same in 
each compound. ‘The spectra obtained from oxides have already been examined.t 
@ 
Spectra of Magnesium, Zine, Cadmium, and Mercury. 
There are points of great interest in the flame-spectra of these metals. 
Lines and bands are present in the spectrum of magnesium burning in air, and 
in the spectra of compounds of this metal heated in the oxyhydrogen flame. The 
* “ Flame-Spectra at High Temperatures,’ Phil. Trans., vol. 185 (A.), 1894, p. 1029. 
t Journ. Chem. Soc., vol. 41, 1882, p. 84. 
{ ‘‘Flame-Spectra at High Temperatures.’ Loe. cit. 
3A 2 
